A two-objective design optimisation approach for blending repairs of damaged compressor blisks

摘要

The blending of compressor blades leads to a permanent modification of the blade geometry. According to these geometric changes, the modal properties of refurbished blades will generally differ from the nominal blade design. Currently, the structural integrity of refurbished blades is maintained by only allowing blends within predefined geometric limits, where the final geometry of the blend is based on a case-by-case decision made by the technician.This work contributes to repair decisions by utilising multi-objective optimisation methods to find structurally optimised blend designs. A parameterised blending model is introduced to uniquely specify the repair design for any blade geometry. Blend designs are analysed systematically by linking the blending model with the Finite Element model of a bladed sector. The influence of the blending shape is evaluated by means of blade-alone frequencies and vibration mode shapes. A two-objective optimisation problem is derived from the frequency tuning of blended blades. The first objective results from the deviation of natural frequencies from the nominal ones, the second objective considers the proximity of natural frequencies to resonance frequencies. Moreover, the approach is applied to one bladed sector of a compressor blisk. The sensitivities of the blend geometry regarding the first six vibration modes are studied. Two damage patterns at different positions at the leading edge of the blade are chosen to exemplarily demonstrate the capability of the proposed approach. For both damage patterns, the set of Pareto optimal solutions is found using Global Pattern Search. The courses of the Pareto frontiers clearly differ and are identified to be specific for each damage pattern. Depending on the damage pattern, discontinuous as well as continuous Pareto frontiers are determined. Providing Pareto optimal solutions for repair designs, the developed approach can be understood as a valuable complement to purely experience-based design decisions. (C) 2020 Elsevier Masson SAS. All rights reserved.